Building intelligence from first principles — where mathematical rigor meets real-world impact. MS Data Science @ Boston University · Open to full-time roles.
I believe the most impactful intelligence is built from the ground up. While many rely on the "magic" of pre-built libraries, my approach to Data Science is rooted in mathematical rigor and fundamental clarity.
"Whether implementing Gaussian Mixture Models from scratch or guiding students through Python at BU — my focus is mastering the fundamentals to build systems with purpose."
My work sits at the intersection of Machine Learning and Large Language Models, driven by a commitment to developing projects that move beyond "proof of concept" into meaningful, real-world utility.
Beyond data, I am a former National Quizzer on Indian National Television — a background that fuels my lifelong pursuit of clarity and rapid knowledge synthesis.
Working under Prof. Elizabeth A. Barnes and Prof. Thomas Gardos at BU's Barnes Lab, contributing to ML-HFI 2.0 — a large-scale deep learning initiative training models on 25TB of Landsat satellite imagery to predict the Human Footprint Index across global land surfaces from 1999–2024. Currently extending the pipeline by developing a delta model to quantify multi-year change in land-use pressure dynamics, prognosing how human activity reshapes ecosystems over time.
Served as Data Scientist and Tech Lead for MassMutual's Data Days for Good initiative, overseeing GitHub-based project coordination and troubleshooting technical challenges. Processed raw LinkedIn education data through Python-based cleaning pipelines, transforming unstructured JSON into meaningful academic insights that informed the New Commonwealth Fund's community programs.
Led an end-to-end ML project focused on predicting employee promotions. Extracted and validated HR data from JSON files using hashing for integrity verification, then engineered a Random Forest classifier achieving 92% accuracy. Expanded analysis by implementing K-means clustering, ANNs, and SVMs to benchmark model performance across different employee profiles.
A multimodal AI comic book generator built solo in two weeks for the Google Cloud × Gemini Hackathon. AbsoluteCinema acts as a cinematic creative director — using Gemini 2.0 Flash on Vertex AI to stream narration and comic panel art together in one interleaved output, layering in Google Cloud TTS voiceovers. Pick your genre, describe your story, and watch a full comic issue materialize panel by panel.
Built at the Vibe Coding Together @ Harvard event, this app compares 15 U.S. cities across six weighted dimensions — Public Transit, Cost of Living, Job Market Density, Weather, Nightlife, and Proximity to Nature. Tailored for tech roles like Data Scientist, ML Engineer, and Software Engineer, it draws live data from Open-Meteo, BLS OEWS, Numbeo, and WalkScore to help professionals decide where to plant their next career flag.
A fiduciary-grade stock risk analysis platform built with LangChain and Chainlit. Integrates real-time financial data, news sentiment, and SEC Edgar 10-K risk factors to deliver comprehensive risk assessments for any publicly traded ticker.
An interactive visualization platform aggregating regional data on income, education, crime, and demographics across New England — helping users make informed relocation decisions.
Combines interactive visualizations of profitability, growth, leverage, and liquidity metrics with a Gradient Boosting model predicting future EPS.
Collaborated with Senator Edward Markey's office to map federal funding equity across 350+ Massachusetts municipalities using Python, Folium, and OpenStreetMap.
Atul A. Das, Muthu Thiruvengadam, Benod Kumar Kondapavuluri, G. R. Brindha
Analyzed careers of Virat Kohli, Jos Buttler, and David Warner using statistical modelling and ML. Applied SVM, Logistic Regression, and Naïve Bayes to predict dismissal outcomes — SVM achieved >80% accuracy, Naïve Bayes reached 94% precision and 92% recall. Linked EDA patterns with predictive modelling to generate actionable insights for players, coaches, and analysts.
Read Paper ↗Atul A. Das, Dr. Sivaprakash JS
Comparative financial analysis of the England and Wales Cricket Board and New Zealand Cricket — examining P&L statements, balance sheets, equity changes, and cash flows to surface trends in revenue, investment, and strategic direction across both boards.
Read Paper ↗Dr. Vidya V (contributing author)
Contributed a chapter to this book on problem-centric innovation — exploring how Design Thinking's empathising phase reframes conventional problem-solving by prioritizing the problem statement over the solution.
View on Amazon ↗A die-hard fan of Manchester United and Bayern Munich — football isn't just something I watch, it's something I live. Most Sundays you'll find me in Swampscott, lacing up my cleats and hitting the pitch.
From the Chanderkhani Pass in India to Blue Mountain in Canton, MA and Valley Forge National Historic Park — hiking offers the quiet I need to reset and recharge.
Former National Quizzer who appeared on Indian National Television in 11th grade. That competitive curiosity still drives everything — from rapidly synthesizing research papers to building better ML intuition.
Active participant in BosMUN and BarMUN at Boston University. MUN sharpens the same skills that make data science compelling — synthesizing complex information and communicating nuanced ideas clearly.
Open to full-time roles in Data Science, ML Engineering, Data Analytics, and Data Engineering.
Based in Boston, MA · OPT authorized · Available now.
Or reach me directly at atularavinddas@gmail.com · +1 (857) 605-9478